Petrologic and textural diversity among the PCA 02 howardite group, one of the largest pieces of the Vestan surface


Corresponding author. E-mail:


Abstract— Nine howardites and two diogenites were recovered from the Pecora Escarpment Icefield (PCA) in 2002. Cosmogenic radionuclide abundances indicate that the samples are paired and that they constituted an approximately 1 m (diameter) meteoroid prior to atmospheric entry. At about 1 m in diameter, the PCA 02 HED group represents one of the largest single pre-atmospheric pieces of the Vestan surface yet described. Mineral and textural variations were measured in six of the PCA 02 howardites to investigate meter-scale diversity of the Vestan surface. Mineral compositions span the range of known eucrite and diogenite compositions. Additional non-diogenitic groups of Mg- and Fe-rich olivine are observed, and are interpreted to have been formed by exogenic contamination and impact melting, respectively. These howardites contain olivine-rich impact melts that likely formed from dunite- and harzburgite-rich target rocks. Containing the first recognized olivine-rich HED impact melts, these samples provide meteoritic evidence that olivine-rich lithologies have been exposed on the surface of Vesta. Finally, we present a new method for mapping distributions of lithologies in howardites using 8 elemental X-ray maps. Proportions of diogenite and eucrite vary considerably among the PCA 02 howardites, suggesting they originated from a heterogeneous portion of the Vestan surface. While whole sample modes are dominated by diogenite, the finer grain size fractions are consistently more eucritic. This discrepancy has implications for near-infrared spectral observations of portions of Vesta’s surface that are similar to the PCA 02 howardites, as the finer grained eucritic material will disproportionately dominate the spectra.


The howardite, eucrite, and diogenite (HED) meteorites comprise the most voluminous group of achondrites in the Antarctic meteorite collection. Spectral evidence strongly suggests that their parent body is the asteroid 4Vesta (McCord et al. 1970), one of the largest differentiated objects in the asteroid belt. Howardites are fragmental and regolith breccias that primarily consist of eucrite (basalt and gabbro) and diogenite (ultramafic cumulate) crystal fragments and clasts. These breccias are formed by impact excavation and mixing on or near the surface of Vesta (Bunch 1975). Howardites typically contain coarse fragments embedded in a fine-grained matrix (e.g., Bunch 1975; Delaney et al. 1984). Petrologic features associated with surface residence (implanted solar wind) and impact (impact melt, exogenic material) are commonly found in howardites due to their formation mechanism. Ground-based telescopic observation suggests that much of the Vestan surface is covered in howardite-like material (Gaffey 1997). Determining the nature and scale of petrologic variations within surface samples is essential for correct geologic interpretations of data returned by Dawn, a NASA mission that arrived at Vesta last year.

The primary instrument on Dawn that will be used to map the surface of Vesta is the visible and near-infrared spectrometer (VIR), which operates between 0.25 and 5.05 μm and has a spatial resolution of approximately 170 m per pixel (Russell et al. 2004). Grain size distribution is important to VIR data interpretation, as absorptions in the spectral range covered by this instrument are dominated by volumetric scattering. Fine-grained textures promote multiple scattering events, resulting in increased reflectance (Hapke 1993). Accordingly, if VIR is used to observe an area where a mixture of coarse and fine grains is present, the finer grained material will contribute disproportionately to the spectral signature. Therefore, it is essential to characterize the grain size distribution of different lithologies expected to be on the Vestan surface for accurate VIR data interpretation.

Along with variance in textures, it is also important to determine localized compositional diversity expected on the Vestan regolith. This will add insight into regolith maturity on Vesta and aid in the interpretation of VIR data as well. Although lithologic distributions with respect to grain size have been studied for individual howardites (e.g., Labotka and Papike 1980; Fuhrman and Papike 1981; Buchanan and Mittlefehldt 2003), variation within a group of paired samples that represent a volumetrically large portion of the Vestan surface has not been investigated.

Nine howardites and two diogenites were collected from the Pecora Escarpment Icefield (PCA) in 2002, and were tentatively paired based on proximity (about 6 km, Fig. 1) and initial petrographic observations (McBride et al. 2004, 2005). In this study, first we measure cosmogenic radionuclides on a subset of this group to confirm pairing and to constrain pre-atmospheric size of the meteoroid. We then measure compositional and textural diversity within the group to ascertain the degree of heterogeneity for the meteoroid, which in turn has implications for heterogeneity on the Vestan surface. We also investigate the distribution of eucrite and diogenite fragments in the howardites with respect to grain size, which has implications for VIR data interpretation.

Figure 1.

 Location of meteorites recovered from the Pecora Escarpment North 40 Icefield during the 2002 field season. All howardites and diogenites were collected within about 6 km. Image courtesy of ANSMET.


Sample Selection

For the purposes of testing the pairing of these samples, we selected three of the nine howardites (PCA 02009, PCA 02013, and PCA 02015) and one of the two diogenites (PCA 02008) for cosmogenic radionuclide analysis. To constrain textural and petrologic diversity among the entire group, we selected six howardite thin sections (PCA 02009,12, PCA 02013,9, PCA 02014,6, PCA 02015,7, PCA 02018,4 and PCA 02019,4) for electron microprobe (EMP) analysis and X-ray mapping. Similar analyses of one of the diogenites, PCA 02008, are available for comparison (Beck and McSween 2010). Additional sections of three meteorites were analyzed to examine smaller, intrasample variation (PCA 02009,7, PCA 02013,6, and PCA 02015,4). The three howardites and one diogenite of the PCA 02 group that were not analyzed by EMP or measured for cosmogenic radionuclides in this study (howardites PCA 02016, PCA 02065, PCA 02066, and diogenite PCA 02017) were examined using a petrographic microscope to confirm their pairing to the rest of the group. For the majority of this article the “PCA 02” prefix will be omitted and samples will be referred to by an abbreviated collection and section number (i.e., PCA 02019,7 = “019,7”).

Chemical and Cosmogenic Radionuclide Analysis

For the analysis of the long-lived cosmogenic radionuclides, we dissolved 50–150 mg samples in a mixture of concentrated HF/HNO3 along with a carrier solution containing about 4 mg of Be and Cl. After complete dissolution, Cl was separated as AgCl and small aliquots of the dissolved samples were taken for bulk chemical analysis by atomic absorption spectrometry (AA) and inductively coupled plasma optical emission spectroscopy (ICP-OES). After measuring the Al concentration in a small aliquot of the dissolved sample, we added 3.5–4.5 mg of Al carrier to the remaining solution. We separated Be and Al using procedures described previously (e.g., Welten et al. 2001) and measured the concentrations of 10Be, 26Al, and 36Cl by accelerator mass spectrometry (AMS) at Purdue University (Sharma et al. 2000). The measured 10Be/Be, 26Al/Al, and 36Cl/Cl ratios were corrected for blanks and normalized to AMS standards (Sharma et al. 1990; Nishiizumi 2004; Nishiizumi et al. 2007).

EMP Analysis and X-Ray Mapping

Mineral analyses were made using a Cameca SX-100 EMP. Analytical conditions were: 20 kV, 30 nA, 2 μm diameter beam for pyroxene, olivine, chromite, metal, and troilite; 15 kV, 10–20 nA, 8–10 μm beam for plagioclase, glass, and phosphate. We analyzed impact melts using a 50–100 μm defocused beam. Counting times for major elements in all analyses ranged from 20 to 40 s, while those for minor elements ranged from 60 to 80 s. Approximately 200 analyses were conducted for each sample. Only analyses that had significant oxide totals (98.5–101 wt%) and acceptable stoichiometry were used.

One component of this work is to constrain eucrite and diogenite grain size variation in these howardites. While cross-polarized images are able to display textural variations, they are limited in their ability to quantify compositional variations that may be tied to textures (Fig. 2a). X-ray maps, however, do show both textural and compositional variability. Typical X-ray map compilations can only represent 3 elemental concentrations at a time, as they are limited to a red–green–blue (RGB) display input. It is possible to add grayscale as a fourth component, but this result is non-unique as pixels with equivalent RGB elemental concentrations will also appear grayscale. These 3-component RGB X-ray maps are ideal for rocks like eucrites that contain a compositionally limited number of phases (e.g., Mayne et al. 2009), but howardites are comprised of a wide range of minerals, so variability is difficult to distinguish with only three or four elements (Fig. 2b). To depict textural variations in these howardites, we devised a method that combines the information from 8 elemental X-ray maps into a single image and uses image-processing software to discreetly map the distribution of specific phase compositions, which correspond to typical HED components (Fig. 2c). We refer to these maps as “lithologic distribution maps” for the remainder of this work. Similar methods have been effectively utilized in other achondrites groups (e.g., Floss et al. 2007), although not as of yet in the HEDs. These lithologic distribution maps not only denote textural variations, but also can be used to determine modal mineralogy. A detailed explanation of the methodology used in creating these maps is given in the Appendix.

Figure 2.

 a) Cross-polarized image of howardite PCA 02015. b) Typical blended X-ray map of 4 elements. c) A false color lithologic distribution map created through the stacking and image processing of 8 elemental X-ray maps where different colors correspond to phases from different HED lithologies and impact induced products. Methods for creating map described in the Appendix.


Mineral Compositions

Pyroxene and Olivine

Pyroxene and olivine compositions are listed in Table 1 and endmember compositions are displayed in Fig. 3. In that figure, diogenite pyroxene (En85–66; Beck and McSween 2010; Shearer et al. 2010; Mittlefehldt et al. 2012), cumulate eucrite pyroxene (En65–46; Mittlefehldt et al. 1998; Mayne et al. 2009), and basaltic eucrite pyroxene (En45–33; Mittlefehldt et al. 1998; Mayne et al. 2009) are denoted by different colors, which are consistent with those used later in the lithologic distribution maps. It is important to note that while the En65–46 and En45–33 ranges generally separate the cumulate and basaltic eucrite pyroxenes, there are some basaltic eucrites that contain pyroxene that falls in the cumulate eucrite range defined here, and vice versa (Mittlefehldt et al. 1998; Mayne et al. 2009). It should be noted that some pyroxenes from the Yamato Type-B diogenites also fall in the cumulate eucrite range defined here (Mittlefehldt and Lindstrom 1993). However, no textural evidence was observed in the PCA 02 howardites that would indicate pyroxenes within that compositional range belong to the Yamato Type-B subgroup of diogenite.

Table 1.   Representative compositions of eucrite and diogenite components.
Oxide wt%
 TiO2 0.080.330.61  0.13bd0.58 
 Al2O3bdd0.810.790.8934.3bd0.780.560.4034.4 0.350.400.9534.4
 Cr2O3bd0.430.270.35 bd0.400.440.20  0.490.210.51 
 MnO0.520.520.730.88 0.520.550.861.09 0.530.400.960.94 
 MgO34.826.920.312.3 34.825.719.211.5 33.528.617.912.40.11
 CaObd1.282.905.9417.70.051.314.440.8817.3 0.441.246.0619.1
 Na2O bdbdbd1.26 bdbdbd1.62 bdbd0.030.86
 K2O    0.08    bd    bd
Cation formula
 Mg1.381.441.140.72 1.381.391.070.69 1.371.521.020.740.01
 Na    0.11    0.14   0.000.08
 K    0.01          
TiO2  0.160.370.52  0.100.390.56 
Al2O3bd0.500.400.5333.6bd0.960.430.5934.6 0.621.150.6134.4
Cr2O3bd0.460.190.28 0.100.700.220.26 0.190.360.890.22 
MnO0.560.540.791.19 0.270.500.900.98 0.430.540.651.00 
MgO37.127.416.912.7 38.225.514.711.8 37.126.620.013.0 
Na2O bdbdbd1.38 bd0.05bd0.66 bd0.030.031.08
K2O    0.08    0.05    0.06
Caption formula
 Mg1.451.470.980.75 1.481.380.850.70 1.461.441.130.77 
 Na    0.12  0.00 0.06
 K    0.01    0.00    0.00
  1. aPyroxene with diogenitic composition (En85–66; Beck and McSween 2010; Shearer et al. 2010; McSween et al. 2012).

  2. bPyroxene with cumulate eucrite composition (En65–46; Mittlefehldt et al. 1998; Mayne et al. 2009; McSween et al. 2012).

  3. cPyroxene with basaltic eucrite composition (En45–33; Mittlefehldt et al. 1998; Mayne et al. 2009; McSween et al. 2012).

  4. dElemental concentration was measured but fell below detection limit.

Oxide weight%
 TiO2  0.12bd0.66 
 Al2O3 0.350.220.2634.80.030.810.490.3634.
 Cr2O30. bd0.570.330.49 bd0.690.290.25 
 MnO0.470.561.030.90 0.520.560.901.04 0.480.500.860.92 
 MgO32.327.616.011.9 35.625.819.711.5 36.525.719.011.4 
 Na2O bdbd0.030.96 bdbdbd1.02 bd0.04bd0.83
 K2O    0.07    0.04    0.04
Caption formula
 Ti  0.00 0.02 
 Mg1.301.480.920.70 1.401.411.110.69 1.441.401.080.68 
 Na    0.09    0.07
 K         0.00    0.00
Figure 3.

 Pyroxene and olivine compositions in the PCA02 howardites. Each quadrilateral/histogram stack represents one section, multiple sections from one sample are grouped. Colored ranges are from Mittlefehldt et al. (1998, 2012), Mayne et al. (2009), Beck and McSween (2010), and Shearer et al. (2010).

All the thin sections contain pyroxenes from each HED sub-lithology (Fig. 3). Most samples display a continuum of compositions across all lithologies; however, both sections of 009 contain three discrete populations of pyroxene corresponding to the diogenite, cumulate eucrite, and basaltic eucrite fields. The majority of pyroxene compositions in each sample are diogenitic, except those in 013. Both sections of 013 have pyroxene compositions that fall primarily in the basaltic eucrite field. Clinopyroxene is observed in all samples, typically as exsolved lamellae associated with cumulate eucrite and basaltic eucrite pyroxenes.

The majority of olivine compositions fall within the diogenite range (Fo78–61; Beck and McSween 2010; Shearer et al. 2010) (Fig. 3). We will refer to olivines within this compositional range as “∼Fo75 olivine” or “diogenitic olivine” for the remainder of this work. All samples also contain olivine in the Fo60–44 range (hereafter “∼Fo50 olivine”), which is more Fe-rich than known diogenite olivine. The Fe/Mn of the ∼Fo50 olivine is 49, approximately the same as diogenitic olivine (Fe/Mn 50; Beck and McSween 2010). PCA 02015 contains the highest concentration of ∼Fo50 olivine, and section ,7 of that sample contains the most Fe-rich (Fo44) olivine of the entire group. PCA 02009, 014, 015, 018, and 019 contain olivine in the Fo88–79 range as well (hereafter “∼Fo85 olivine”), which is more Mg-rich than the diogenites. The average Fe/Mn of the ∼Fo85 olivine is 41, less than typically observed in diogenites. Both ∼Fo50 and ∼Fo85 olivines have been documented in other howardites (Delaney et al. 1980; Fuhrman and Papike 1981), yet their origin remains unclear.

Plagioclase, Silica, and K-Rich Glass, Opaques

Plagioclase compositions are listed in Table 1. In all the PCA 02 howardites, plagioclase falls within the compositional range of eucrites (Fig. 4) (An96–75; McSween et al. 2012). Samples 018, 019, and 015,7 have higher concentrations of calcic plagioclase, consisting of ∼An96–90 plagioclase exclusively. The rest of the samples have a broader range of plagioclase compositions. Glass is primarily silica-rich, ranging from 99.4 to 88.9 wt% SiO2, with the majority being approximately 98 wt% SiO2 (Table 2).

Figure 4.

 Plagioclase compositions in the PCA 02 howardites. Each plot represents one section, multiple sections from one sample are grouped. All compositions fall within known HED ranges (An96–75, McSween et al. 2012).

Table 2.   Representative melt and glass composition.
PhaseOl-rich impact meltOl-poor impact meltSi-rich glassK-rich glass
  1. aMeasured but fell below detection limit.

Oxide wt%

PCA 02014 contains one grain of K-rich glass (10.1 wt% K2O; Table 2). This grain is about 225 μm, angular, almost completely glass, and is mantled by small (10 μm) ∼Fo50 olivines. Compositionally, this glass is similar to the “felsite glass”Barrat et al. (2009) reported in howardite Northwest Africa (NWA) 1664. The K-rich glass in PCA 02014 contains approximately 2 and 4 times the concentrations of K2O and Na2O, respectively, than the felsite glass in NWA 1664. Along with Na2O and K2O, the K-rich glass in 014 is also enriched in P2O5 and Al2O3 and depleted in TiO2, MgO, CaO, and FeO, relative to the NWA 1664 felsite glass (Table 2). PCA 02014 K-rich glass is enclosed within a large olivine-rich impact melt breccia clast. The interpretation of these clasts as impact melts will be addressed below.

FeNi metal ranges from kamacite to taenite, up to 54.2 wt% Ni. Variations in FeNi metal observed in three samples (009,7, 013,6, 015,4; Beck et al. 2007) are consistent with the variations in the rest of the group. The majority of the FeNi metal grains have Co/Ni compositions typical for pallasites, iron meteorites, and chondrites, not HEDs (Beck et al. 2007). This likely indicates that the metal in the PCA 02 samples was intermixed through impact and brecciation of meteoritic debris on the Vestan surface (e.g., Hewins 1979). All sulfides in these samples are troilite. Spinels range from Cr2O3 = 58.5 wt%, TiO2 = 0.63 wt%, Al2O3 = 6.70 wt% (consistent with diogenitic chromite; Mittlefehldt 1994; Bowman et al. 1999) to Cr2O3 = 45.6 wt%, TiO2 = 5.50 wt% (consistent with eucritic ulvospinel; Mayne et al. 2009). Ilmenite compositions also fall within the eucrite range, spanning from 3.2–0.70 wt% MgO and 1.2–0.04 wt% Cr2O3 (Mayne et al. 2009).

Bulk Chemistries and Cosmogenic Radionuclide Abundances

Chemical Composition

Bulk chemistries of the PCA 02 howardites are heterogeneous, and vary by factors of 2–3 in Al and Ca (Table 3). Although 008 was initially classified as a diogenite, some basaltic (eucritic) clasts have been recognized in the sample (Beck and McSween 2010). As shown by Mittlefehldt et al. (2012) and supported here, 008 has higher Al and Ca concentrations than typical diogenites (008 = 1.29 wt% Al, 1.67 wt% Ca; diogenite = 0.90 wt% Al, 1.5 wt% Ca; Mittlefehldt et al. 2012). The Al, Ca, and Mg concentrations of howardite 009,13 are similar to those of diogenite 008, suggesting that 009 contains a high concentration of diogenitic material, as suggested by Warren et al. (2009). The rest of the PCA 02 howardites have higher concentrations of Al and Ca, and lower concentrations of Mg, suggesting that 008 and 009 are diogenite-rich endmembers of the PCA 02 howardite-diogenite group.

Table 3.   Concentrations of major elements in PCA 02 howardites. All values are in wt%, except for K and Co, which are given in ppm. Concentrations of Mg, Al, K, Ca, Mn, and Fe were measured by AA, Ti, and Co by ICP-OES and O and Si were estimated from other elements.
  1. aThe composition of PCA 02009,5 is from Warren et al. (2009).


Bulk siderophile element concentrations in the PCA 02 howardites vary considerably (Table 3). PCA 02009 is depleted in siderophiles (0.02 wt% Ni, 28 ppm Co), while 013 and 015 are enriched (0.11 and 0.32 wt% Ni, 87 and 141 ppm Co, respectively). The Ni and Co concentration in 015 are among the highest reported in howardites, and are about 6 and 4 times higher than the averages for siderophile-rich “regolithic” howardites (Warren et al. 2009). Only howardite Mount Pratt (PRA) 04401, which contains abundant chondritic fragments, has higher Ni concentrations (0.44 wt% Ni; Mittlefehldt et al. 2011) than PCA 02015. The high Ni and Co concentrations in 013 and 015 are consistent with the observation of abundant FeNi metal in the PCA 02 howardite group.

Cosmogenic Radionuclides

The 10Be concentrations of the PCA 02 samples are remarkably constant at 24.0 ± 0.5 dpm kg−1 (Table 4), consistent with the hypothesis that they are paired fragments of a single meteoroid that broke up during atmospheric entry. The 26Al concentrations show slightly more variations, from 80–90 dpm kg−1, with an average of 85 ± 4 dpm kg−1. Part of the observed variations in 26Al can be explained by variations in chemical composition, since the concentrations of Si and Al, the major target elements for the production of 26Al, vary significantly among the four PCA meteorites (Table 3). After normalizing the measured 26Al concentrations to the total concentration of [Si + 1.8*Al], reflecting the approximately 80% higher production rate of 26Al from Al relative to Si (Leya and Masarik 2009), the normalized 26Al concentrations (in units of dpm kg−1 [Si + 1.8*Al]) range from 305 to 329 with an average of 314 ± 10, consistent with the hypothesis that these samples came from a single meteoroid. In contrast to the relatively constant 10Be and 26Al concentrations, the 36Cl concentrations vary by more than a factor of 2, from 7.6 dpm kg−1 in 008 to 17.1 dpm kg−1 in 015. Although it is tempting to conclude that these large variations in 36Cl indicate these meteorites represent different falls, Fig. 5 shows that they can mainly be attributed to variations in the major target elements for 36Cl production, Fe and Ca (Table 3), if a production rate ratio of P(36Cl)Ca/P(36Cl)Fe of 12 ± 2 is assumed, consistent with irradiation at 15–20 cm depth in an object with R = 60 cm. The slope of the correlation in Fig. 5 yields a constant 36Cl concentration of 23.0 ± 1.1 dpm kg−1 [Fe + 12Ca], indicating that the PCA howardites came from a single shower, which fell relatively recently, i.e., less than 50 kyr ago.

Table 4.   Concentrations of cosmogenic radionuclides (in dpm/kg) of PCA 02 HEDs.
PCAType 10Be 26Al 26Ala 36Cl 36Cla
  1. aBulk 26Al concentrations are normalized to Si + 1.8Al, those of 36Cl are normalized to Fe + 12Ca.

02008Dio24.4 ± 0.586.4 ± 1.9329 ± 77.6 ± 0.222.6 ± 0.6
02009How23.4 ± 0.580.8 ± 2.1305 ± 89.2 ± 0.322.2 ± 0.7
02013How24.3 ± 0.584.4 ± 2.0311 ± 811.0 ± 0.322.2 ± 0.6
02015How23.6 ± 0.589.5 ± 2.3312 ± 817.1 ± 0.624.5 ± 0.8
Figure 5.

 Dependence of bulk 36Cl (in dpm kg−1) on the concentrations of major target elements, Fe and Ca, in the PCA 02 howardites and PCA 02008 diogenite, assuming that the relative production rate from Ca is 12 ± 2 times higher than from Fe. The solid line corresponds to an average 36Cl content of 23 dpm kg−1 [Fe + 12Ca], while the dashed lines show the uncertainty in the relative production of 36Cl from Ca.

The high 10Be and 26Al concentrations in the PCA howardites constrain the pre-atmospheric radius of the object to 90–230 g cm−2, whereas the 36Cl concentration in 015 yields a somewhat tighter range of 120 180 g cm−2 (Fig. 6). Assuming an average density of 3.0 g cm−3 for howardites (Britt and Consolmagno 2003), these values correspond to absolute radii of 40–60 cm, which are consistent with the radius of 50–60 cm derived from the high contribution of neutron-capture produced 41Ca in 008 (Welten et al. 2007).

Figure 6.

 Comparison of measured concentrations (represented by the gray bars) of cosmogenic 10Be (a), 26Al (b), and 36Cl (c, d) in PCA howardites and PCA 02008 diogenite with calculated production rates (dashed and solid curves) as a function of depth in howardites with pre-atmospheric radii of 35–300 g cm−2 (roughly 12–100 cm). The production rates are based on the elemental production rates in ordinary chondrites (Leya and Masarik 2009) and the chemical composition of the PCA samples from Table 2. The gray bars in figure a and b represent the average 10Be and 26Al concentrations in the four PCA samples; the 26Al concentrations and production rates are normalized to the total concentration of [Si + 1.8*Al] to correct for differences in chemical composition between the samples. For 36Cl, we compared the measured concentration in the sample with the lowest Ca content (02008; c) and highest Ca content (02015; d) with calculated production rates. The elemental production rates of 36Cl from Ca given by Leya and Masarik (2009) were increased by 20% to match the observed 36Cl concentrations in PCA 02008 and 02015.


Classification of Secondary Material

The PCA 02 howardite thin sections are comprised mostly of eucrite and diogenite clasts and mineral fragments. However, there are a few other petrologic components that were likely formed through secondary processes on the surface of Vesta. Before discussing the distribution of primary eucritic and diogenitic material, we first categorize the secondary components.

Eucrite breccia clasts. PCA 02013,9 and 019 contain clasts that are dominated by basaltic eucrite crystal fragments enclosed in a fine-grained matrix (Fig. 7a). The eucrite crystal fragments and lithic clasts are about 75 and approximately 500 μm, respectively, where distinct crystal shapes and primary igneous grain boundaries can be seen. The matrix separating these eucritic clasts is too fine to resolve (less than 2 μm), but has high concentrations of opaque minerals and possibly eucritic material (Fig. 7b). The bulk composition of the matrix is slightly less ferroan and more aluminous than is typical of eucrites (fine-grained matrix = 48.6 wt% SiO2, 15.6 wt% Al2O3, 15.6 wt% MgO, 8.2 wt% CaO, 10.7 wt% FeO versus bulk eucrite = 48.6 wt% SiO2, 12.2 wt% Al2O3, 8.9 wt% MgO, 10.4 wt% CaO, 18.8 wt% FeO; Kitts and Lodders 1998). The fine-grained matrix is enriched in SO2 (0.80 wt%) and NiO (0.03 wt%), likely due to the high concentration of opaque phases. An approximately 1 mm vitrophyric fragment with spherical opaque inclusions is also found within this clast (Fig. 7c). The bulk composition of this vitrophyric fragment is approximately the same as the matrix chemistry listed above.

Figure 7.

 BSE image of a eucrite breccia clast in PCA 02019,4. a) This clast is comprised of eucritic fragments containing plagioclase + pyroxene + silica, b) which are separated by a fine-grained matrix with high concentrations of opaques and eucritic material. c) This clast also contains an opaque-bearing vitrophyric fragment. This fragment is approximately eucritic in composition, and is likely either a quenched impact melt, or piece of lithified regolith.

These eucrite breccia clasts are “breccias-within-breccias,” indicating that the PCA 02 howardites contain materials representing multiple impact events (Pun et al. 1998). Given the composition of these eucrite breccia clasts (dominantly eucritic w/sub-micrometer scale metal and sulfide blebs) and their textures (very fine grained), it is possible that these clasts represent older, more extensively reworked Vestan surface material that has been lithified and incorporated into these relatively younger and fragmental PCA 02 howardite breccias. Conversely, these may be eucrite impact melt breccia clasts that were incorporated into the PCA 02 howardites sometime after their formation.

Olivine-rich impact melts. Most of the PCA 02 howardites contain clasts that can be best described as having an olivine-phyric basaltic texture. These clasts contain 0.2–1 mm olivine phenocrysts set in a fine-grain groundmass (about 10 μm crystals) of olivine (∼Fo50), > pyroxene (∼En65), > plagioclase, >> opaques (mainly FeNi metal and sulfide) (Figs. 8 and 9a–c). We interpret these clasts as impact melt breccia clasts (Bischoff et al. 2006), where phenocrysts and lithic fragments are partially resorbed pieces of eucrite and diogenite and the matrix is quenched impact melt. Olivine phenocrysts in these impact melt breccia clasts have ∼Fo75 core compositions, are partially resorbed, and zoned to ∼Fo50 at their rims, where they are in equilibrium with the matrix (Figs. 8b and 9a). Core compositions of phenocrysts fall in the diogenite range (diogenites = Fo78–61; Beck and McSween 2010; Shearer et al. 2010) while rims do not. Also present are partially resorbed lithic clasts composed of ∼Fo75 olivine and ∼En75 orthopyroxene (Fig. 8), which are similar in texture and composition to the harzburgitic diogenites (Beck and McSween 2010). There are also large (about 300 μm) partially resorbed phenocrysts of En75 orthopyroxene, which are diogenitic in composition and texture (McSween et al. 2012). As with larger olivine grains, orthopyroxene phenocrysts are zoned and in equilibrium with the melt matrix at their rims (Fig. 8b). A limited number of partially resorbed phenocrysts of plagioclase and more Fe-rich orthopyroxene/pigeonite occur, and they are eucritic in composition. While the melt matrix seen in Fig. 8 only contains 8 vol%∼Fo50 olivine, most melts have higher ∼Fo50 olivine abundances (e.g., Fig. 9b, 47 vol%∼Fo50 olivine). The ranges in olivine abundances in the matrix correspond to ranges in impact melt compositions. As discussed in the Methods section, impact melt compositions were measured using defocused beam analyses on portions of the melt matrix in impact melt breccia clasts. Impact melts with high olivine abundances are enriched in MgO and FeO, and depleted in SiO2, Al2O3, CaO, TiO2, and Na2O, relative to those with less olivine (Table 2). In most clasts, small (less than 5 μm) spherules of troilite and metal occur in the impact melt matrix (Fig. 9c). Note that the K-rich glass discussed above is enclosed in an impact melt clast and can be seen in Fig. 9a.

Figure 8.

 Impact melt breccia clast in PCA 02015. a) BSE image and b) lithologic distribution map of the same area. The impact melt is predominantly ∼En65 Opx and plagioclase, and ∼Fo50 olivine. Larger, partially resorbed clasts of harzburgitic (Hzbg) diogenite, diogenitic olivine (∼Fo75), and diogenitic Opx (∼En75) can also be seen. These partially resorbed clasts and phenocrysts have equilibrated with the impact melt, as seen by their Fe-enrichment from core to rim. The composition of the melt matrix in this impact melt breccia clasts corresponds to the “Ol-poor impact melt” composition in Table 2. Note that the BSE image in (a) was taken under different conditions than those shown in Figs. 7 and 9, causing common phases between the figures having different shades of gray.

Figure 9.

 BSE images of an impact melt breccia clast in the howardite PCA 02014,6. a) Clasts are composed of large, partially resorbed phenocrysts of ∼Fo75 olivine, zoned to ∼Fo50 at rims. Larger partially resorbed Opx phenocrysts, and one K-rich glass, are also observed. b) Impact melt matrix composed mainly of ∼Fo50 olivine, with lesser amounts of pyroxene and plagioclase. The mode of (b) is ol 47%, pyx 32%, plag 20%, metal+sulfide 1%, and a more typical mode for the impact melts than shown in Fig. 8. c) Some impact melt contain higher concentrations of sulfide and or metal spherules. The composition of the melt matrix in this impact melt breccia clasts broadly corresponds to the “Ol-rich impact melt” composition in Table 2.

We interpret these as impact melts rather than volcanic basalts primarily based on textures and compositional properties (i.e., partially resorbed grains, fine-grained textures, and high concentrations of siderophile elements). These indicators have been used to distinguish impact melt rocks in lunar samples (Lucey et al. 2006), and in other howardites (Hewins and Klein 1978). Furthermore, the chemistry of these clasts would be difficult to explain petrogenetically if interpreted as erupted eucritic basalts (i.e., high concentration of olivine is inconsistent with the silica-normative and generally less mafic chemistry of eucrites). Finally, evidence suggests that eucrites and diogenites were not closely related temporally or spatially during their formation (McSween et al. 2012). If these clasts were eucritic basalts, the included partially resorbed diogenitic clasts and crystal fragments would suggest a closer petrogenetic relationship between eucrites and diogenites than previously thought.

Chondritic fragment. A possible chondrule was identified in 014 (Fig. 10). This is the only possible chondritic fragment observed in the nine howardite sections. This circular fragment is 400 μm and composed primarily of En65 orthopyroxene with small amounts of plagioclase, and spherules of metal and troilite that appear to be concentrated toward the rim. Orthopyroxene Fe/Mn is 33 and is approximately in the range of HED low-Ca pyroxene (28–32; Papike et al. 2003); it also falls in the range of chondritic low-Ca pyroxene (4–68; Brearley and Jones 1998), although the chondritic range is much broader due to lack of differentiation. If this fragment is chondritic, chemistry and mineralogy suggest it would be a type IIB chondrule (McSween 1977). If this fragment is not exogenic in origin, based on its texture and composition, it would likely be a diogenitic impact melt clast.

Figure 10.

 A possible type IIB chondrule in PCA 02014, comprised primarily of pyroxene (Wo2 En64 Fs34), with small amounts of interstitial plagioclase, metal (mt), and sulfide (sf).

Lithologic Distribution Maps

Samples with two thin sections. We have separated the samples into two groups for the basis of presenting the lithologic distribution maps. The eight maps have been combined into two figures for printing purposes, and individual high-resolution maps of each sample can be found in the supplementary online material for this article. Modal mineralogy was calculated from the lithologic distribution maps and is given in Table 5. The first group of maps, shown in Fig. 11, consists of samples where two thin sections of the same meteorite were mapped. We attempted to make maps of both sections of PCA 02015; however, one section contained too high a concentration of impact melt and could not be properly mapped. Included in Fig. 11 is a key and % mode histogram for each section. Note that the colors used to denote eucrite breccia clasts and plagioclase are similar (white and pink, respectively). Eucrite breccia clasts can be distinguished by their larger size (about 1 mm), included fragments of eucritic material, and exclusive occurrence in 013,9 and 019,4.

Table 5.   Modal mineralogy.
Diogenitic materialVolume%
 Olivine (Fo80–60)
 Opx (En85–66)69.188.859.933.637.135.336.473.9
 Eucritic material        
 Pyx (cumulate, En65–46)
 Pyx (basaltic, En45–33)12.52.412.621.611.79.717.44.9
 Silica glass0.
Secondary material        
 ≥Fo81 olivine0.
 ≤Fo59 olivine0.
 Impact melt breccia3.
 Eucrite breccia clasts0.
 Metal and sulfide0.
Figure 11.

 Lithologic distribution maps of two howardites. Two sections from each meteorite are shown. Colors represent minerals from HED lithologies. Phase distributions shown on histograms and in Table 5. Individual maps available in supplementary online material. Note: “Eucrite breccia” (white) only occurs in 013,9. Black/gray = background.

A large variation in lithologic abundances is observed between the two sections of 013. Section ,6 is dominated by coarse diogenitic orthopyroxene and olivine (62 vol% combined, typically ≥ 500 μm, some up to 5 mm), with lesser amounts of fine-grain basaltic and cumulate eucritic pyroxene, plagioclase, and glass (35 vol% combined, typically ≤ 75 μm). For the remainder of this discussion “diogenitic material” will refer to diogenite orthopyroxene + ∼Fo75 olivine + chromite, while “eucritic material” will refer to cumulate and basaltic eucrite pyroxene (low and high-Ca) + plagioclase + ilmenite + phosphate + silica glass. Section, 6 of 013 contains an approximately 1 mm impact melt breccia clast with an enclosed harzburgitic diogenite fragment (013,6; Fig. 11). Section ,9 of 013 is composed of 34 vol% diogenitic material, while eucritic material comprises 59 vol% of the sample. As with the other section of 013, diogenitic material in ,9 is larger (120–200 μm) than the eucritic material (50–125 μm). PCA 02013,9 also contains a large, approximately 1.2 mm eucrite breccia clast (013,9; Fig. 11). The two sections of 009 display more similarities than the sections of 013 discussed above. Both sections of 009 are dominated by large (≥2 mm) fragments of diogenitic material (,7 = 70 vol%, ,12 = 91 vol%), which are separated by a fine-grained matrix (≤80 μm, with the majority being about 40 μm) dominated by eucritic material (,7 = 26 vol% and ,12 = 8 vol%). A few approximately 400 μm eucrite fragments occur in 009,7. Olivine in 009,7 is small (80 μm), and occurs throughout the eucrite-dominated matrix, except in impact melt breccia clasts, where olivine is slightly larger (150 μm). Section ,7 also contains a 125 μm phosphate grain (magenta) on the right side of the map (009,7; Fig. 11). Section ,12 of 009 contains no impact melt breccia clasts, and olivine in this sample is mainly found in distinct olivine + orthopyroxene, harzburgitic diogenite clasts. The cumulate eucrite pyroxene in this sample is restricted in distribution to a single fragmental vein between two large diogenite orthopyroxene fragments on the right side of the sample (009,12; Fig. 11). While the sum of diogenitic versus eucritic material in 009,12 (91 versus 8 vol%) suggests that it is a “polymict diogenite,” other sections of this sample suggest that “howardite” is a more appropriate taxonomic term (009,7 = 70 vol% diogenite versus 26 vol% eucrite) (Delaney et al. 1984). This contradiction in thin section modes presents challenges to the proper classification of this sample, something which we will not address herein. To avoid confusion we will continue to refer to all sections of 009 as “howardite,” following the initial classification of that sample.

Samples with one thin section. The lithologic distribution maps for the single-section samples are displayed in Fig. 12. PCA 02018 contains the highest concentration of metal + sulfide in the group (8 vol%), and also appears to be the most metal-rich howardite reported to date. The breakdown of this metal is roughly 70% kamacite and 30% taenite (up to 55 wt% Ni). The metal is concentrated in an approximately 3 mm2 area and is associated with very fine-grained, melt-like textures (018; Fig. 12). Given the abundance, composition, and textural setting of this metal, it was most likely introduced by meteoroid impact. PCA 02018 has approximately equal proportions of diogenitic and eucritic material (40 and 46 vol%, respectively), but these lithologies vary in grain size. Diogenitic orthopyroxene and olivine are typically large (≥400 μm), whereas eucrite pyroxene and plagioclase are small (≤70 μm) and in the matrix. This sample also contains two elongated basaltic eucrite clasts, which preserve igneous contacts between enclosed phases. There are two clasts of harzburgitic diogenite in this section: a 1 mm grain in the center of the sample, and a 400 μm grain enclosed in the impact melt breccia clast in the upper right portion of the sample (018,4; Fig. 12).

Figure 12.

 Lithologic distribution maps for the remaining howardites. See Fig. 11 caption for key and details. Note: “Eucrite breccia” (white) only occurs in 019,4.

PCA 02014 has the second highest abundance of impact melt (20 vol%), which occurs in impact melt breccia clasts that also contain large olivine and orthopyroxene phenocrysts, and a few about 250 μm pieces of metal + sulfide (014,6; Fig. 12). The impact melt breccia clasts can be large, measuring up to approximately 4 mm2. The amount of diogenitic versus eucritic material in 014 is roughly equal (43 versus 36 vol%). However, the average grain size of diogenitic material is much larger, typically ranging from 400 μm to 1.1 mm, with a few ≥ 2 mm clasts, while most of the eucritic material is ≤70 μm (014,6; Fig. 12). There are three large (1–2 mm) cumulate eucrite clasts in this section, which occur in the top right, center, and bottom middle portions of the sections.

PCA 02015,7 contains the highest concentration of impact melt (28 vol%), which like 014 above, occurs in impact melt breccia clasts containing olivine and pyroxene phenocrysts, along with some metal grains (015,6; Fig. 12). This sample consists of about 40 vol% diogenitic material, while eucritic material only comprises 29 vol% of the sample. As with previous samples, diogenitic material is much coarser (most grains are ≥ 200 μm, with the majority 600 μm), while eucrite grain fragments are smaller (typically ≤ 90 μm). A single 1 mm cumulate eucrite clast occurs in the top portion of this section, and an approximately 800 μm basaltic eucrite clast is in the bottom left portion of the section (015,6; Fig. 12). This sample also contains two harzburgitic diogenite fragments; one on the left half of the section, the other is elongated and on the right.

PCA 02019 has the second highest concentration of diogenitic pyroxene (74 vol%), which are typically 1.5 mm with some mineral clasts up to 4 mm. Olivine (4 vol%) occurs as 300 μm grains in the matrix, and are in high concentrations on the left side of the section, enclosed in a 3 mm harzburgitic diogenite clast (019,3; Fig. 12). The majority of basaltic and cumulate eucrite pyroxene, plagioclase, and silica grains (combined 17 vol%) are ≤ 100 μm and dispersed in the matrix between larger diogenite clasts and grains. Some eucrite clasts are as large as 1 mm. A 1.5 mm eucrite breccia clast is found in the upper right portion of this sample. It contains crystal and lithic fragments of basaltic eucrite material. Impact melt breccia clasts in 019 are ≤ 1 mm.



As discussed above, cosmogenic radionuclides abundances confirm that diogenite 008, and howardites 009, 013, and 015 are paired. The three howardites of that group are dominated by coarse-grained diogenitic clasts and contain olivine-rich impact melt breccia clasts (Figs. 11 and 12). Two of these samples, 009 and 015, contain both Fo- and Fa-rich olivines, which fall outside the diogenite range (Fig. 3). The three howardites that were analyzed in this study by EMP, but not measured for cosmogenic radionuclides (014, 018, and 019) share each of these distinct petrologic characteristics, indicating that they are part of the larger pairing group as well.

We investigated the petrography of the remaining PCA 02 HEDs (howardites 016, 065, 066, and diogenite 017) to test their pairing with this group. Howardites 066,4 and 016,4 contain impact melts with olivine phenocrysts, fragments of harzburgitic diogenite, in some cases relatively high abundances of metal (estimated 2 vol%), and are generally dominated by diogenitic components. This supports their paring with the rest if the PCA 02 howardite-diogenites group. Howardite 065,3 consists primarily of mm-sized orthopyroxenitic diogenite clasts set in a fine-grained matrix of ≤ 50 μm fragments of orthopyroxene, with minor amounts of clinopyroxene and plagioclase. The petrography of that sample is very similar to that of the howardite 009 described in this study, and similar to that of the diogenite 008 described by Beck and McSween (2010), both of which have been paired to the PCA 02 group. Therefore, howardite 065 likely also belongs to the pairing group as well. Diogenite 017,2 is different. It is composed of smaller, approximately 500 μm orthopyroxene grains, set in a fine-grained, ≤70 μm, orthopyroxene-dominated matrix with minor amounts of olivine. Given the mineralogy of this sample (diogenite-rich) and its proximity of recovery to the other samples (Fig. 1), it too is likely paired with the group, but it is texturally unique.

Mineral Compositions


Two olivine compositions that are found in the PCA 02 howardites (∼Fo50 and ∼Fo85) fall outside of the diogenitic olivine range (Fo78–61; Beck and McSween 2010; Shearer et al. 2010). It has been hypothesized that Mg-rich olivines in other howardites, up to Fo96, were derived from a yet to be discovered Mg-rich diogenitic lithology (Delaney et al. 1980). This is one possibility for the ∼Fo85 olivines we observe in the PCA 02 howardites. However, if the ∼Fo85 grains were incorporated from the brecciation and mixing of an Mg-rich diogenitic lithology, we would expect to find associated fragments of this lithology or other phases from this lithology (i.e., ∼En87 that is in equilibrium with ∼Fo85 olivine). We do not find any of these fragments. Two recent studies involving olivine in diogenites, which examined 13 of the 16 olivine-bearing diogenites in the Antarctic meteorite collection, did not reveal any ≥ Fo80 olivine (Beck and McSween 2010; Shearer et al. 2010), further suggesting ∼Fo85 olivines in the PCA 02 howardites did not originate from a diogenitic lithology. Furthermore, the Fe/Mn of the ∼Fo85 olivine is approximately 40, which is lower than that of diogenitic olivine Fe/Mn (about 50; Beck and McSween 2010), suggesting the ∼Fo85 olivine did not originate from Vesta.

Another interpretation is that the ∼Fo85 olivine is exogenic, and that it was incorporated into the howardites from the impact mixing of foreign material on the surface of Vesta. While the Fe/Mn of the ∼Fo85 olivine is lower than the HEDs, it is very similar to that in the main-group pallasites (Fe/Mn = 40; Mittlefehldt et al. 1998). Similarly, Mg concentrations of the ∼Fo85 olivine are greater than those in the diogenites, but in the range of those in the main-group pallasites (Fo82–89; Mittlefehldt et al. 1998). A pallasitic origin for the ∼Fo85 olivine is further supported by composition and abundance of metal in the PCA 02 howardites. As discussed above and in Beck et al. (2007), a large portion of the metal in these samples has Co/Ni concentrations in the pallasite, iron meteorite, chondrite range. The abundance of metal observed here (up to 8 vol%) suggests contamination from a meteorite rich in metal, like the pallasites (about 15–65 vol% metal; Mittlefehldt et al. 1998). We favor the interpretation that ∼Fo85 olivine in the PCA 02 howardites was derived from an exogenic, likely pallasitic source.

The ∼Fo50 olivine in the PCA 02 howardites has a more straightforward origin. As was discussed previously, most of the PCA 02 samples contain impact melt breccia clasts. We find ∼Fo50 olivine almost exclusively as small (about 10 μm), subhedral crystals in the melt matrix of those clasts (Figs. 8 and 9). This indicates that the ∼Fo50 olivine originated from the quenching of impact melts. This is supported by Fo75–50 zoning (core to rim) of partially resorbed olivine pheoncrysts (Fig. 9), indicating that they equilibrated with a melt capable of producing ∼Fo50 olivine. It is important to point out the distinction that being formed by impact melt, the ∼Fo50 olivines in these howardites were formed by secondary processes on the Vestan surface and did not crystallize from diogenitic magmas or eucritic basalts.

Olivine-Rich Impact Melts

Target Rock Petrology

Given the high abundances of olivine phenocrysts, partially resorbed olivine-rich lithic fragments, and olivine grains in the melt matricies of the PCA 02 howardite impact melt breccia clasts (Figs. 8 and 9), the target rock for these impact melts was likely olivine-rich. This is further supported by the compositions of the impact melts, which are ultramafic (Table 2). The partially resorbed lithic fragments in the melt breccia clasts are texturally and compositionally similar to harzburgitic diogenites (Beck and McSween 2010), suggesting that lithology may have been the target rock. However, large (≥ 2 mm) euhedral olivine phenocrysts with no included or adjacent orthopyroxene, also occur in PCA 02 impact melt breccias (Fig. 9a). Olivine grains of this size, which also lack associated orthopyroxene, are rare to non-existent in harzburgitic diogenites (Beck and McSween 2010), although grains with similar textures have been noted in the Miller Range (MIL) 03443 dunite of the HED group (Beck et al. 2011). Therefore, it is plausible that the target rock for these impact melts may have contained a dunitic component as well. Given that dunitic and harzburgitic diogenites are likely co-genetic on Vesta (Beck et al. 2011), and therefore may occur in close proximity, both lithologies could be sampled and incorporated into an impact melt during a single impact event.

The occurrences of plagioclase, along with relatively Fe-rich pyroxene (∼En65) and olivine (∼Fo50) in the melt matrices of these impact melt breccia clasts (Figs. 8 and 9) suggest that the target rock was not solely comprised of dunitic and harzburgitic diogenite. Harzburgitic and dunitic diogenites contain less than 1 vol% plagioclase and have ≥ Fo71 olivine and comparable pyroxene compositions (Bowman et al. 1997; Beck and McSween 2010; Beck et al. 2011). An impact melt that was derived exclusively from these two diogenitic lithologies would likely quench to form phases with similar compositions and abundances, not minerals enriched in Fe, Al, and Ca. Similarly, the bulk compositions of PCA 02 impact melts are calcium- and aluminum-rich relative to harzburgitic and dunitic diogenite bulk compositions (PCA 02 impact melts = 4.5–10.2 wt% Al2O3, 3.2–7.4 wt% CaO, harzburgite = 0.9 wt% Al2O3, 1.4 wt% CaO; Mittlefehldt et al. 2012; dunite = 0.5 wt% Al2O3, 0.1 wt% CaO; Beck et al. 2011).

We propose that the target rock for the PCA 02 impact melts also contained a eucritic component, which when melted with harzburgitic and dunitic diogenite components, would add sufficient Al, Ca, and Fe to the melt to crystallize plagioclase and Fe-rich olivines and pyroxenes. The few partially resorbed eucritic crystal and lithic fragments in some impact melt breccia clasts support this hypothesis (Figs. 11 and 12). This suggests that the target rock for the PCA 02 impact melts was howarditic (eucrite + diogenite) with diogenite portions that were primarily dunitic and harzburgitic (Ol-rich). Therefore, a generalized geologic history for the formation of these melts would be as follows: initial impact events that excavate and incorporate harzburgitic and dunitic material into the eucritic crust (formed the olivine-rich howardite), and a subsequent impact event which melts the combined material.

Proportion of Olivine-Rich Components

Although it is difficult to quantify the abundances of olivine-rich diogenitic versus eucritic components in the original target rock, we can estimate their relative proportions by examining the abundances of undigested diogenitic and eucritic lithic clasts and mineral fragments in the impact melt breccia clasts. As we have already discussed, undigested clasts in these impact melts are primarily harzburgitic and dunitic (Figs. 8, 9, 11, and 12), suggesting that these lithologies were the primary constituents of the target rock. We now examine the compositions of the PCA 02 impact melts to further test that hypothesis.

PCA 02 impact melt compositions are compared to eucrite and diogenite bulk compositions in Fig. 13 using a modified mixing diagram after Usui and McSween (2007). PCA 02 impact melt compositions plot primarily along tie-lines between eucrite and harzburgitic or dunitic diogenite. The location of the PCA 02 impact melt compositions along these tie-lines suggest that the target material ranged from about 25 to 75% dunitic and harzburgitic diogenite, though some compositions fall outside this range. Two variables lead us to conclude that the target rocks were more dunite- and harzburgite-rich than the approximately 25–75% range. First, the data plotted in Fig. 13 are compositions of the melt matrix in the impact melt breccia clasts, and it is likely that the more evolved eucritic portions of the target rock (Si-, Ca-rich/Fe-, Mg-poor) preferentially melted relative to the ultramafic diogenitic portions, and thus are oversampled in melt matrix compositions. Likewise, since the values plotted in Fig. 13 are matrix compositions and do not incorporate the large undigested phenocrysts of olivine and harzburgite clasts, they are not true representations of the bulk chemistry (and thus petrology) of the target rock. If undigested clasts and mineral fragments were incorporated into these calculations, more Fe-, Mg-rich compositions would be achieved. Finally, the measurements in Fig. 13 were made using a defocused electron beam, and due to density contrasts between olivine and plagioclase, the actual concentrations of Fe and Mg in the matrix are likely higher than are reported in Fig. 13 (Warren 1997). While ascertaining true target rock chemistry would require bulk chemical or modal recombination analysis of these impact melt breccia clasts, the above lines of evidence strongly suggest that the dunitic and harzburgitic portions of the target rock were greater than the 25–75 vol% range estimated from melt matrix compositions, supporting the hypothesis that the target rock was a howardite primarily composed of dunitic and harzburgitic diogenite.

Figure 13.

 PCA 02 impact melts compared to eucrite and diogenite compositions, after Usui and McSween (2007). Separate diogenite lithologies are distinguished and connected to eucrites via tie-lines. Percent diogenitic component along these tie-lines has been noted. Previously published HED impact melt compositions are also plotted, and overlay the tie-line between eucrite and orthopyroxenitic (opxn) diogenite. Most PCA 02 impact melts, however, suggest mixing of eucritic and harzburgitic or dunitic (Ol-rich) diogenite. 1. Desnoyers and Jerome (1977), 2. Metzler and Stöffler (1987, 1995), 3. Metzler et al. (1995), 4. Pun et al. (1998), 5. Buchanan and Mittlefehldt (2003), 6. Barrat et al. (2009), 7. Kitts and Lodders (1998), 8. Beck et al. (2011), 9. Irving et al. (2003), 10. Mittlefehldt et al. (2012), 11. Barrat et al. (2008).


Note that along with PCA 02 impact melts, compositions from impact melts in other HEDs are plotted in Fig. 13 as well. All previous HED impact melt compositions plot on, or at the approximate endmembers of, an orthopyroxenitic (Ol-free) diogenite and eucrite tie-line. However, as discussed above, the PCA 02 impact melts are uniquely derived from an olivine-rich protolith. The identification of olivine-rich HED impact melts has implications for the geologic history of Vesta, interpretation of data from NASA’s Dawn mission, and future petrology studies of HED meteorites.

Impact melt breccias form in the highest pressure regimes (75–90 GPa; Bischoff and Stöffler 1992) during high-velocity impacts on asteroid bodies (4.5–5 km/s; Bischoff et al. 2006), which indicates that they originate in a transient melt zone on, or directly beneath, the surface during cratering events (Bischoff and Stöffler 1992). Therefore, impact melts are a good indication of a lithology’s exposure on the surface of an asteroid. Although some spectral evidence suggests olivine-rich lithologies may be exposed on the Vestan surface (e.g., Gaffey 1997), no evidence has yet been identified in the HED meteorites supporting this hypothesis. The olivine-rich impact melts in the PCA 02 howardites, however, provide the first meteoritic evidence that olivine-rich lithologies have been exposed on the surface of Vesta at some point during its history.

These impact melts resemble the textures and chemistries that would be expected for olivine-rich eucritic basalts (∼Fo50 olivine + Fe-rich pigeonite + plagioclase, quenched textures). It is possible that Dawn may observe areas on the surface of Vesta that have similar compositions and textures, and it is also possible that similar fragments will be observed in other HED meteorites. If similar lithologies occur in larger areas that can be resolved by Dawn on the surface of Vesta or are observed in other meteorite samples, the interpretation that they are impact melts and not igneous olivine-rich basalts would be supported. These melts also provide additional evidence that dunites occur on Vesta. MIL 03443 is the only dunite that has been tied to the HEDs (Beck et al. 2011). With the observation of olivine mineral fragments in the PCA 02 howardite impact melt breccia clasts that resemble those in MIL 03443, and impact melt compositions suggesting a dunite-bearing target rock, the occurrence of dunites on Vesta is further supported.

Finally, impact melts generated from large impact events, such as the impact that formed the approximately 500 km wide Rheasilvia basin on the south pole of Vesta (Schenk et al. 2012), would crystallize from a melt sheet, and thus would be close to uniform in composition. The olivine-rich impact melts in the PCA 02 howardites are clearly nonuniform in composition (Table 2; Fig. 13), implying that these melts formed from separate, smaller impact events, and not from the Rheasilvia event. Because the target rock was howarditic, and the impact event that excavated the olivine-rich diogenite material happened at some point prior to melting, discussing the implications that melt compositions might have on the distribution of olivine-rich diogenitic lithologies on Vesta would be inappropriate (i.e., the nonuniformity in melt compositions does not necessitate small, separate impacts to excavate the original igneous, olivine-rich material).

Lithologic Distribution Maps

Lithologic Heterogeneity: Small and Large-Scale Variation

Although this work was not singularly driven toward characterizing the scale of heterogeneity in howardite samples, several trends that may be useful for constraining surface variability on Vesta are revealed. By examining the variation in lithologic components between multiple sections of one sample (intrasample variation), we can infer variation at an approximately centimeter scale. Between sections ,6 and ,9 of 013, the concentrations of diogenitic and eucritic material vary by a factor of approximately 2 (Table 5). Section ,6 contains 62% diogenitic and 35% eucritic materials, whereas section ,9 contains about 34% diogenitic and 59% eucritic materials. The two sections of 009 vary by similar amounts, with diogenitic and eucritic material varying in abundance by factors of about 0.2 and 2, respectively, between the two sections. Similar intrasample variations also occur with the abundances of impact melt in these two samples (e.g., 009,12 = 0 vol% versus 009,5 = 4 vol%). These results suggest that within a centimeter-sized scale on the surface of Vesta the amount of eucritic, diogenitic, and secondary material varies significantly.

Examining variation among the entire group of paired samples (intersample variation) may constrain larger, meter-scale variations on the Vestan surface. Among the entire group of samples we see larger variance. Diogenitic material varies by a factor of about 3 across the nine sections (34–91 vol%), while eucritic material varies by a factor of about 7 (8–59 vol%). Secondary components vary in similar proportions. Metal + sulfide range from 0 to 8 vol%, and impact melt abundance ranges from 0 to 28 vol%. The large lithologic variability of these samples indicates that the portion of the surface where they originated is very heterogeneous, and likely immature (i.e., had not been extensively reworked by impacts and thus homogenized).

There has been a recent increase in attempts to identify true regolithic howardites, or those samples that resided on the outer most portion of the Vestan surface (e.g., Warren et al. 2009). Although the true test for regolithic howardites typically involves solar noble gas concentrations, several petrologic indicators denoting regolithic residence, including abundances of metal (or high siderophile element concentrations), and impact glass, have also been proposed (Warren et al. 2009). Herein we demonstrate that within a 1 m area on the surface of Vesta both metal and impact glass abundances range from 0 vol% to the highest concentrations that have been reported in howardites (8 and 28 vol%, respectively). Furthermore, we also demonstrate that the abundances of siderophile elements measured in bulk samples vary by similar degrees within the paired meteorites (0.02–0.32 wt% Ni). These variations suggest that metal and impact melt abundances have not been thoroughly mixed in the Vestan regolith, and may be enriched in localized areas. This implies that the abundances of metal and impact glass may not be characteristic of all truly regolithic howardites and may explain some of the disconnect between siderophile element abundances and solar noble gas concentrations (Cartwright et al. 2011).

Lithologic Grain Size Distribution: Implications for VIR

As seen in the lithologic distribution map of each sample, diogenitic material dominates the coarser grain size fractions. Typically, diogenitic material is ≥ 500 μm, but in two samples it is much larger, nearer 1.5 mm. Individual mineral grains of eucritic material, however, are ≤ 70 μm in all samples. This variance in grain sizes between the lithologies is likely partially due to the contrasting grain sizes of the original material (i.e., diogenites are coarse-grained plutonic rocks versus fine-grained eucritic basalts). However, it may also be due to differing residence times on the surface of Vesta between the diogenitic and eucritic material that was incorporated into these breccias (i.e., finer grained eucritic material was exposed on the surface and reworked longer than diogenitic material). This suggests that the diogenitic material in these breccias was excavated and incorporated into the surface later than the eucritic portions.

In this study we demonstrate that eucritic components typically only make up about 30 vol%, but dominate the fine-grain size fraction of these howardites. Similar findings have been demonstrated in other howardites (e.g., Labotka and Papike 1980; Buchanan and Mittlefehldt 2003), suggesting that this phenomenon may not be localized. This order-of-magnitude grain size difference between eucritic and diogenitic material could be problematic for VIR spectral interpretation. Each time light passes between two grains, there is the possibility of scattering at the interface. Other investigators (e.g., Hapke 1993) have shown that the trend of decreasing grain size amplifies the reflectance of the sample in the visible and near-infrared wavelength range. Consequently, this phenomenon will result in increasing the overall contribution of the fine-grained eucritic material (versus the coarser-grained diogenitic material) to a reflectance spectrum of the PCA 02 samples. In other words, the eucritic material will disproportionately dominate a VIR spectrum of surface regolith, if the lithologic grain size distribution we observe here is common.

Grain size effects, as well as other factors, in the visible and near-infrared wavelength range complicate spectral deconvolution or the estimation of modal abundances of natural surfaces that are mixtures of multiple phases (such as regolith). Therefore, characterizing possible regolith grain size variability in howardites can be used to help constrain different VIR data reduction techniques. The models of Hapke (1981, 1993) and Shkuratov et al. (1999) assume an average particle radius or the distance light propagates through a particle before internal reflection. These parameters, and the derived grain sizes modeled by these two approaches, can be evaluated for consistency with the lithologic grain size distribution from this study.


  • 1 The PCA 02 howardites contain silicate mineral compositions that span almost all known HED ranges, suggesting that their source region on Vesta is diverse, and likely not thoroughly mixed (i.e., immature).
  • 2 Two groups of olivines in the PCA 02 howardites have compositions outside those known for diogenites. One group (∼Fo85) is more Mg-rich than diogenites, which is likely due to exogenic contamination. The other, more Fe-rich group (∼Fo50) crystallized from impact melts.
  • 3 Olivine-rich impact melts comprise up to 28 vol% of the PCA 02 howardites. The target rocks for these melts were howarditic (diogenite + eucrite), but the diogenite component was olivine-rich (dunitic or harzburgitic diogenite) and was a chief component of the target material. These are the first olivine-rich impact melts identified in HEDs, and provide the first meteoritical evidence that olivine-rich lithologies have been exposed on the surface of Vesta.
  • 4 Lithologic variation observed among PCA 02 howardites suggests that portions of the Vestan surface might show significant m-scale lithologic variations. Eucritic material is fine-grained, whereas diogenitic material is very coarse. This will likely affect the interpretation of VIR spectra. These data might be used to test models that correctly estimate the proportions of these two lithologies in the Vestan regolith.

Acknowledgments— The authors thank the MWG and ANSMET for the collection and allocation of meteorite samples. This work was partially supported by NASA Cosmochemistry Grant NNG06GG36G and UCLA Dawn Team subcontract to HYM. We thank D. Mittlefehldt and A. Patzer for their constructive comments. We also thank MAPS AE A. Ruzicka and referees J.-A. Barrat and P. Buchanan for their helpful reviews and editorial handling.

Editorial Handling–– Dr. Alex Ruzicka


Appendix. Lithologic Distribution Map Methods

To create lithologic distribution maps, we first acquired eight elemental X-ray images of a sample on the electron microprobe. The eight elements measured were Mg, Al, Si, K, Ca, Ti, Cr, and Fe. We used a 2 μm beam set at 20 kV and 30 nA, an 8 μm step size, and 50 ms counting times for each element. A 16 bit image range was necessary to preserve each approximately 0–2000 counts possible in each elemental X-ray image. Each of the eight-element X-ray maps of a given sample were mosaiced and stacked into a single image cube using the image-processing software Environment for Visualizing Images (ENVI 4.2), where each band of the cube represents a single element. Background epoxy and cracks in the thin section were masked and removed using a threshold of counts below 340 for all elements.

A region of interest (ROI) was made for each of the typical phases distinguished in this study. In all, 17 ROIs were defined: diogenite olivine (Fo78–61), diogenite pyroxene (En85–66), cumulate eucrite pyroxene (En65–46), basaltic eucrite pyroxene (En45–33), plagioclase, clinopyroxene, silica glass, chromite, ilmenite, phosphate, ∼Fo85 olivine, ∼Fo50 olivine, FeNi metal, sulfide, phosphate, impact melt, and eucrite breccia clasts. This is a robust list of all phases expected in an HED meteorite (Mittlefehldt et al. 1998) and all were measured through detailed spot analysis in our samples. To create each ROI, we chose a select number of grains with a known compositional range (from our EMP data) that fell within our 17 classifications. For example, for the cumulate eucrite pyroxene ROI, we chose several pyroxene grains with compositions between En65–46, taking care to select a uniform distribution of En values within the range. Plagioclase, silica, clinopyroxene, the opaque phases, phosphate, eucrite breccia clasts, and impact melt were relatively homogeneous in composition within each sample, and therefore easier to define with an ROI.

After creating the ROIs, we applied a minimum distance supervised classification technique (ENVI 4.2) to classify each pixel in our image cube as one of the 17 classifications derived from our ROIs. The minimum distance classification calculates the Euclidean distance of each unknown pixel to the center of each ROI class. Pixels are classified to the closest ROI class, unless they fall outside the defined standard deviations for each class, in which case they are “unclassified.” We assigned different maximum values for the standard deviation from the mean of each ROI to produce results that most closely resembled observations of these samples in thin section. They were Fo78–61 = 12, En85–66 = 15, En65–46 = 5.5, En45–33 = 11, plagioclase = 13.5, clinopyroxene = 7, silica glass = 14, chromite = 8, ilmenite = 8, phosphate = 10, ∼Fo85 olivine = 8, ∼Fo50 olivine = 10 metal = 10, sulfide = 10, fine-grained eucrite breccia = 2.5, and impact melt = 2.5. The minimum distance used for fine-grained eucrite breccia and impact melt are particularly low, because if they were set higher this technique would misclassify pixels from the very fine-grained parts of the matrix as either one of these two ROIs. A small portion of fusion crust in 013,9 was misclassified as impact melt. This did not significantly alter the mode.

Each sample had approximately ≤10% unclassified pixels. As stated above, we believe a large amount of the unclassified pixels are very fine grains in the matrix that were too small to be resolved by the 8 μm step size. In order to estimate approximate error, we randomly selected 15 compositional analyses collected by EMP in each section and checked their compositions measured by EMP versus the ENVI assigned classification. The classification and EMP compositions agreed approximately 95% of the time. The only phases that fell below 95% were clinopyroxene, impact melt, and eucrite breccia, and their maximum standard deviations used in ROI classification were lowered accordingly, which decreased the approximate error. Note that some grains classified as En86–66, which is dark teal, display non-uniform color distributions; some pixels within these grains are classified as En65–46 and are aquamarine. This is primarily an indication of compositional variation within single grains (i.e., zoning from an Mg-rich to Fe-rich pyroxene). However, in a few instances this is the result of analyses falling near fractures, resulting in lower X-ray intensities, and leading to pixels being misclassified as the more Mg-depleted (En65–46) phase.